Constructing Valid Geospatial Tools for Environmental Justice (2024) / Chapter Skim
Currently Skimming:
6 Indicator Integration
6 Indicator Integration
Pages 137-166
The Chapter Skim interface presents what we've algorithmically identified as the most significant single chunk of text within every page in the chapter.
Select key terms on the right to highlight them within pages of the chapter.
From page 137... ...
Chapter 3 provided discussion on the construction of composite indicators. This chapter discusses those process components in the context of CEJST and their effect on the designation of census tracts as disadvantaged communities.
|
From page 138... ...
In the sections below, the major decisions associated with indicator integration are discussed in detail. BRINGING INDICATORS ONTO A COMMON SCALE The goal of calculating a composite indicator is to combine different data types and measurements into a single measure or indicator to obtain as complete a picture of a complex, multidimensional phenomenon as possible (Mazziotta and Pareto, 2017)
|
From page 139... ...
For example, CEJST uses the 90th percentile threshold on the indicator of low median income to determine if a census tract is designated as disadvantaged in the workforce development category. By this logic, a census tract at the 90th percentile is not considered any less disadvantaged than a census tract at the 99th percentile.
|
From page 140... ...
Mathematically, weights are typically applied by simply multiplying an indicator before aggregating it with other weighted indicators, thus increasing or decreasing the relative influence of each indicator on the resulting composite measure. There is an art and a science to creating composite indicators and weighting.
|
From page 141... ...
Many composite indicators do not have explicit weights assigned to each indicator (Freudenberg, 2003) , and few of the environmental justice tools described in Chapter 4 use an explicit weighting scheme.
|
From page 142... ...
If a census tract meets the criteria for be ing designated as disadvantaged based on indicators in a single burden category, it would have a "total categories exceeded" value of 1. If a census tract met the criteria based on indicators in all categories, it would have a value of 8.
|
From page 143... ...
. the output composite indicator best reflects the concept being measured, alternative approaches to using nominal categories could be considered.
|
From page 144... ...
. For example, if both diabetes and heart disease indicators are included in a composite indicator and are not combined into a subindex, this dimension of health would have a greater influence on the resulting composite indicator due to implicit weighting.
|
From page 145... ...
Based on the methodology currently employed in CEJST, the categories of climate change, housing, and legacy pollution have 2.5 times the share of indicators used to designate a census tract as disadvantaged. This is not inherently right or wrong, but understanding the statistical impact of these decisions and justifying them is crucial to ensure that the resulting composite indicator is representative of the concept being measured.
|
From page 146... ...
The documentation also does not explicitly discuss the purpose of the burden categories, whether organizational or methodological. Overall, indicator weights have significant impacts on the resulting composite indicator.
|
From page 147... ...
states that "disadvantaged communities face numerous challenges because they have been marginalized by society, overburdened by pollution, and underserved by infrastructure and other key services," the implementation of CEJST stops short of reflecting those numerous challenges in their designation of census tracts as disadvantaged. While each census tract has multiple opportunities to be considered disadvantaged based on the 30 individual input indicators (described in Chapter 5)
|
From page 148... ...
. As opposed to the binary classification of indicator criteria or burden categories exceeded in CEJST, these aggregation approaches generate continuous values.
|
From page 149... ...
In the additive approach, despite Community B having two indicators with significantly lower values, the higher value for Indicator 3 compensates for the low values and leads to an index value in the middle. With the multiplicative approach, the two low indicators in Community B lower the overall index value substantially.
|
From page 150... ...
150 FIGURE 6.1 Frequency distribution of the number of census tracts, with cumulative impacts alternatively represented by the count of indicator threshold criteria exceeded (top) , and the count of burden categories exceeded (bottom)
|
From page 151... ...
FIGURE 6.1 Continued 151
|
From page 152... ...
The final composite indicator values are based on alternative aggregation approaches (sum and product) to the census tracts designated as disadvantaged based on the health category within the current CEJST implementation.
|
From page 153... ...
Diabetes Asthma Heart Disease Expectancy Low Income Sum Product (max = 4) Louisiana, Orleans 0.99 0.97 0.99 0.93 0.99 4.87 0.88 4 Parish (22071014000)
|
From page 154... ...
CalEnviroScreen is an environmental justice screening tool that uses this approach. After multiplicative aggregation, it designates tracts scoring in the top 25th percentile as disadvantaged.4 Any resulting composite indicator will be sensitive to choices made during normalization, weighting, aggregation, and post-processing, including the interactions between them.
|
From page 155... ...
Another objective and important benefit of sensitivity analysis is that it can help differentiate the influence that input parameter options have on the composite indicator (e.g., those that greatly influence DAC designation and those that do not)
|
From page 156... ...
Global sensitivity analysis evaluates the response of an output composite indicator to simultaneous variations among multiple modeling parameters. Un certainty analysis is first done to determine the probabilistic distribution of the composite indicator, typically using Monte Carlo simulation.
|
From page 157... ...
In these instances, the ability to simultaneously vary multiple input parameters would be useful to interrogate the robustness of the current CEJST configuration, as well as potential future changes to CEJST, such as adding new variables or considering alternative percentile thresholds. Global uncertainty and sensitivity analysis are better suited to evaluate the sensitivity of a model to uncertainties in multiple input parameters.
|
From page 158... ...
Monte Carlo simulation is employed to construct the model repeatedly, with each iteration generating the model output based on a random selection of the input parameter options. Instead of a discrete output value for each analysis unit (e.g., census tracts)
|
From page 159... ...
This example demonstrates the utility of uncertainty analysis for both quantifying overall composite indicator fragility to alternative modeling decisions and identifying which analysis units have the most and least reliable ranks. For CEJST, the large number of census tracts (>80,000)
|
From page 160... ...
Sensitivity analysis is thus applied to decompose the overall variance and assign proportions to individual input modeling parameters. In this fashion, the analyst could determine, hypothetically, that the choice of environmental burden threshold is responsible for 35 percent of the variation in percent DAC designation, while the normalization parameter is responsible for only 5 percent.
|
From page 161... ...
. Unlike local sensitivity analysis, global sensitivity analysis can distinguish between main and interaction effects.
|
From page 162... ...
Low High Low Additive Robust Use the model as is or optionally reduce epistemic uncertainty for input parameters with high main effects to further increase robustness. Low Low High Interactive Robust Use the model as is or optionally reduce epistemic uncertainty for input parameters with high interactive effects to further increase robustness.
|
From page 163... ...
Box 6.5 provides an example workflow for uncertainty and sensitivity analysis that might be useful to CEQ. Sensitivity analysis techniques are important tools for improving the robustness of models, increasing the transparency of the model construction, and ultimately increasing the validity of the model.
|
From page 164... ...
of interest and the objec tives of the sensitivity analysis.
|
From page 165... ...
Modelers select input sample distributions for the modeling choices of interest, conduct a Monte Carlo simulation, extract outputs of the resulting uncertainty distribution, and apply sensitivity analysis to quantify the relative influence of each modeling choice.
|
From page 166... ...
CEQ could conduct uncertainty and sensitivity analysis on CEJST to quantify the influence of input composite indicator modeling decisions on the output designation of disadvantaged communities.
|
Key Terms
This material may be derived from roughly machine-read images, and so is provided only to facilitate research.
More
information on Chapter Skim is available.